From 527c46b595024d9fe5f9da95b32b25c26b6bcb02 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 08:45:04 -0700
Subject: [PATCH 01/26] add manual deprecation into validation

---
 emmet-core/emmet/core/vasp/validation.py | 39 ++++++++++++++++--------
 1 file changed, 27 insertions(+), 12 deletions(-)

diff --git a/emmet-core/emmet/core/vasp/validation.py b/emmet-core/emmet/core/vasp/validation.py
index 9de9997a73..f4961dc6fd 100644
--- a/emmet-core/emmet/core/vasp/validation.py
+++ b/emmet-core/emmet/core/vasp/validation.py
@@ -13,6 +13,7 @@
 
 class DeprecationMessage(DocEnum):
     MANUAL = "M", "manual deprecation"
+    DEPRECATED_TAGS = "M001", "Deprecated tag"
     KPTS = "C001", "Too few KPoints"
     KSPACING = "C002", "KSpacing not high enough"
     ENCUT = "C002", "ENCUT too low"
@@ -56,6 +57,7 @@ def from_task_doc(
         input_sets: Dict[str, PyObject] = SETTINGS.VASP_DEFAULT_INPUT_SETS,
         LDAU_fields: List[str] = SETTINGS.VASP_CHECKED_LDAU_FIELDS,
         max_allowed_scf_gradient: float = SETTINGS.VASP_MAX_SCF_GRADIENT,
+        deprecated_tags: Optional[List[str]] = None,
     ) -> "ValidationDoc":
         """
         Determines if a calculation is valid based on expected input parameters from a pymatgen inputset
@@ -73,12 +75,18 @@ def from_task_doc(
         structure = task_doc.output.structure
         calc_type = task_doc.calc_type
         inputs = task_doc.orig_inputs
+        bandgap = task_doc.output.bandgap
 
         reasons = []
         data = {}
 
         if str(calc_type) in input_sets:
-            valid_input_set = input_sets[str(calc_type)](structure)
+
+            # Ensure inputsets that need the bandgap get it
+            try:
+                valid_input_set = input_sets[str(calc_type)](structure, bandgap=bandgap)
+            except TypeError:
+                valid_input_set = input_sets[str(calc_type)](structure)
 
             # Checking K-Points
             # Calculations that use KSPACING will not have a .kpoints attr
@@ -137,18 +145,25 @@ def from_task_doc(
                 ):
                     reasons.append(DeprecationMessage.LDAU)
 
-            # Check the max upwards SCF step
-            skip = inputs.get("incar", {}).get("NLEMDL")
-            energies = [
-                d["e_fr_energy"]
-                for d in task_doc.calcs_reversed[0]["output"]["ionic_steps"][-1][
-                    "electronic_steps"
-                ]
+        # Check the max upwards SCF step
+        skip = inputs.get("incar", {}).get("NLEMDL")
+        energies = [
+            d["e_fr_energy"]
+            for d in task_doc.calcs_reversed[0]["output"]["ionic_steps"][-1][
+                "electronic_steps"
             ]
-            max_gradient = np.max(np.gradient(energies)[skip:])
-            data["max_gradient"] = max_gradient
-            if max_gradient > max_allowed_scf_gradient:
-                reasons.append(DeprecationMessage.MAX_SCF)
+        ]
+        max_gradient = np.max(np.gradient(energies)[skip:])
+        data["max_gradient"] = max_gradient
+        if max_gradient > max_allowed_scf_gradient:
+            reasons.append(DeprecationMessage.MAX_SCF)
+
+        # Check for Manual deprecations
+        if deprecated_tags is not None:
+            bad_tags = list(set(task_doc.tags).intersection(deprecated_tags))
+            if len(bad_tags) > 0:
+                reasons.append(DeprecationMessage.DEPRECATED_TAGS)
+                data["bad_tags"] = bad_tags
 
         doc = ValidationDoc(
             task_id=task_doc.task_id,

From 6a5daf49ea7e239a1bb7ac47581ee535ead1a1a1 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 08:53:29 -0700
Subject: [PATCH 02/26] switch default deprecated tags to setting

---
 emmet-core/emmet/core/settings.py        | 4 ++++
 emmet-core/emmet/core/vasp/validation.py | 4 ++--
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/emmet-core/emmet/core/settings.py b/emmet-core/emmet/core/settings.py
index 23fd23b494..60cfbeba60 100644
--- a/emmet-core/emmet/core/settings.py
+++ b/emmet-core/emmet/core/settings.py
@@ -79,6 +79,10 @@ class EmmetSettings(BaseSettings):
         description="Maximum upward gradient in the last SCF for any VASP calculation",
     )
 
+    VASP_DEPRECATED_TAGS: List[str] = Field(
+        [], description="List of tags to manually deprecate in task validation"
+    )
+
     class Config:
         env_prefix = "emmet_"
         extra = "ignore"
diff --git a/emmet-core/emmet/core/vasp/validation.py b/emmet-core/emmet/core/vasp/validation.py
index f4961dc6fd..d019e66a04 100644
--- a/emmet-core/emmet/core/vasp/validation.py
+++ b/emmet-core/emmet/core/vasp/validation.py
@@ -1,5 +1,5 @@
 from datetime import datetime
-from typing import Dict, List, Union
+from typing import Dict, List, Optional, Union
 
 import numpy as np
 from pydantic import BaseModel, Field, PyObject
@@ -57,7 +57,7 @@ def from_task_doc(
         input_sets: Dict[str, PyObject] = SETTINGS.VASP_DEFAULT_INPUT_SETS,
         LDAU_fields: List[str] = SETTINGS.VASP_CHECKED_LDAU_FIELDS,
         max_allowed_scf_gradient: float = SETTINGS.VASP_MAX_SCF_GRADIENT,
-        deprecated_tags: Optional[List[str]] = None,
+        deprecated_tags: List[str] = SETTINGS.VASP_DEPRECATED_TAGS,
     ) -> "ValidationDoc":
         """
         Determines if a calculation is valid based on expected input parameters from a pymatgen inputset

From fc8a94ca6fc7b90ad1276005f57d0f6279061428 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 09:05:54 -0700
Subject: [PATCH 03/26] move use_statics to settings

---
 emmet-core/emmet/core/settings.py      | 5 +++++
 emmet-core/emmet/core/vasp/material.py | 4 ++--
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/emmet-core/emmet/core/settings.py b/emmet-core/emmet/core/settings.py
index 60cfbeba60..4a57b36967 100644
--- a/emmet-core/emmet/core/settings.py
+++ b/emmet-core/emmet/core/settings.py
@@ -83,6 +83,11 @@ class EmmetSettings(BaseSettings):
         [], description="List of tags to manually deprecate in task validation"
     )
 
+    VASP_USE_STATICS: bool = Field(
+        True,
+        description="Use static calculations for structure and energy along with structure optimizations",
+    )
+
     class Config:
         env_prefix = "emmet_"
         extra = "ignore"
diff --git a/emmet-core/emmet/core/vasp/material.py b/emmet-core/emmet/core/vasp/material.py
index 43cd95f92c..862c628e84 100644
--- a/emmet-core/emmet/core/vasp/material.py
+++ b/emmet-core/emmet/core/vasp/material.py
@@ -44,8 +44,8 @@ class MaterialsDoc(CoreMaterialsDoc, StructureMetadata):
     def from_tasks(
         cls,
         task_group: List[TaskDocument],
-        quality_scores=SETTINGS.VASP_QUALITY_SCORES,
-        use_statics: bool = False,
+        quality_scores: Dict[str, int] = SETTINGS.VASP_QUALITY_SCORES,
+        use_statics: bool = SETTINGS.VASP_USE_STATICS,
     ) -> "MaterialsDoc":
         """
         Converts a group of tasks into one material

From f929bf9df1be6514dc8c05df3da32a43d1c63cf3 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 09:06:57 -0700
Subject: [PATCH 04/26] clean up imports

---
 emmet-core/emmet/core/settings.py      | 6 +++---
 emmet-core/emmet/core/vasp/material.py | 9 +++------
 2 files changed, 6 insertions(+), 9 deletions(-)

diff --git a/emmet-core/emmet/core/settings.py b/emmet-core/emmet/core/settings.py
index 4a57b36967..4f096f0f84 100644
--- a/emmet-core/emmet/core/settings.py
+++ b/emmet-core/emmet/core/settings.py
@@ -1,13 +1,13 @@
 """
 Settings for defaults in the core definitions of Materials Project Documents
 """
-import importlib
 import json
-from typing import Dict, List, Optional, Type, TypeVar, Union
+from pathlib import Path
+from typing import Dict, List, Type, TypeVar, Union
 
 import requests
 from pydantic import BaseSettings, Field, root_validator, validator
-from pydantic.types import Path, PyObject
+from pydantic.types import PyObject
 
 DEFAULT_CONFIG_FILE_PATH = str(Path.home().joinpath(".emmet.json"))
 
diff --git a/emmet-core/emmet/core/vasp/material.py b/emmet-core/emmet/core/vasp/material.py
index 862c628e84..bca02a369c 100644
--- a/emmet-core/emmet/core/vasp/material.py
+++ b/emmet-core/emmet/core/vasp/material.py
@@ -1,12 +1,9 @@
 """ Core definition of a Materials Document """
-from datetime import datetime
-from functools import partial
-from typing import ClassVar, List, Mapping, Optional, Sequence, Tuple, TypeVar, Union
+from typing import Dict, List, Mapping, Sequence
 
-from pydantic import BaseModel, Field, create_model
+from pydantic import Field
 from pymatgen.analysis.structure_analyzer import SpacegroupAnalyzer
-from pymatgen.analysis.structure_matcher import ElementComparator, StructureMatcher
-from pymatgen.core import Structure
+from pymatgen.analysis.structure_matcher import StructureMatcher
 from pymatgen.entries.computed_entries import ComputedStructureEntry
 
 from emmet.core import SETTINGS

From a590c4e8a7dd21f04451d252ddb579d1de390d40 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 09:56:47 -0700
Subject: [PATCH 05/26] add warnings for kspacing and ismear

---
 emmet-core/emmet/core/vasp/validation.py | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/emmet-core/emmet/core/vasp/validation.py b/emmet-core/emmet/core/vasp/validation.py
index d019e66a04..ab3d46d795 100644
--- a/emmet-core/emmet/core/vasp/validation.py
+++ b/emmet-core/emmet/core/vasp/validation.py
@@ -1,5 +1,5 @@
 from datetime import datetime
-from typing import Dict, List, Optional, Union
+from typing import Dict, List, Union
 
 import numpy as np
 from pydantic import BaseModel, Field, PyObject
@@ -79,6 +79,7 @@ def from_task_doc(
 
         reasons = []
         data = {}
+        warnings = []
 
         if str(calc_type) in input_sets:
 
@@ -110,6 +111,15 @@ def from_task_doc(
                     # larger KSPACING means fewer k-points
                     if data["kspacing_delta"] > kspacing_tolerance:
                         reasons.append(DeprecationMessage.KSPACING)
+                    elif data["kspacing_delta"] < kspacing_tolerance:
+                        warnings.append(
+                            f"KSPACING is lower than input set: {data['kspacing_delta']}"
+                            " lower than {kspacing_tolerance} ",
+                        )
+
+            # warn, but don't invalidate if wrong ISMEAR
+            if inputs.get("incar", {}).get("ISMEAR") > 0 and bandgap > 0:
+                warnings.append("Inappropriate smearing settings")
 
             # Checking ENCUT
             encut = inputs.get("incar", {}).get("ENCUT")
@@ -172,6 +182,7 @@ def from_task_doc(
             valid=len(reasons) == 0,
             reasons=reasons,
             data=data,
+            warnings=warnings,
         )
         return doc
 

From cfc7176d3e45725a3818eef2c02e1395d090b270 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 10:36:17 -0700
Subject: [PATCH 06/26] update test since default for use_statics is now True

---
 tests/emmet-core/vasp/test_materials.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/emmet-core/vasp/test_materials.py b/tests/emmet-core/vasp/test_materials.py
index 73f9992076..33222945bf 100644
--- a/tests/emmet-core/vasp/test_materials.py
+++ b/tests/emmet-core/vasp/test_materials.py
@@ -29,7 +29,7 @@ def test_make_mat(test_tasks):
     ]
 
     with pytest.raises(Exception):
-        MaterialsDoc.from_tasks(bad_task_group)
+        MaterialsDoc.from_tasks(bad_task_group, use_statics=False)
 
 
 def test_schema():

From 0210fc207d862fc8639c3bbda8ed541b0c928220 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 12:16:02 -0700
Subject: [PATCH 07/26] speed up PD construction

---
 emmet-core/emmet/core/thermo.py | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/emmet-core/emmet/core/thermo.py b/emmet-core/emmet/core/thermo.py
index 056d14911e..c547ca0023 100644
--- a/emmet-core/emmet/core/thermo.py
+++ b/emmet-core/emmet/core/thermo.py
@@ -1,17 +1,13 @@
 """ Core definition of a Thermo Document """
-from datetime import datetime
-from enum import Enum
+from collections import defaultdict
 from typing import ClassVar, Dict, List, Union
 
 from pydantic import BaseModel, Field
-from pymatgen.analysis.phase_diagram import PhaseDiagram, PhaseDiagramError
-from pymatgen.core import Composition
-from pymatgen.core.periodic_table import Element
+from pymatgen.analysis.phase_diagram import PhaseDiagram
 from pymatgen.entries.computed_entries import ComputedEntry, ComputedStructureEntry
 
 from emmet.core.material_property import PropertyDoc
 from emmet.core.mpid import MPID
-from emmet.core.structure import StructureMetadata
 
 
 class DecompositionProduct(BaseModel):
@@ -94,7 +90,17 @@ class ThermoDoc(PropertyDoc):
     @classmethod
     def from_entries(cls, entries: List[Union[ComputedEntry, ComputedStructureEntry]]):
 
-        pd = PhaseDiagram(entries)
+        entries_by_comp = defaultdict(list)
+        for e in entries:
+            entries_by_comp[e.composition.reduced_formula].append(e)
+
+        # Only use lowest entry per composition to speed up QHull in Phase Diagram
+        reduced_entries = [
+            sorted(comp_entries, key=lambda e: e.energy_per_atom)[0]
+            for comp_entries in entries_by_comp.values()
+        ]
+
+        pd = PhaseDiagram(reduced_entries)
 
         docs = []
 

From 85850771bfbe5a3b7d60cb4a0d082b6c2acb1da3 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 14:15:34 -0700
Subject: [PATCH 08/26] remove deprecation from core since its in builders

---
 emmet-core/emmet/core/settings.py        | 4 ----
 emmet-core/emmet/core/vasp/validation.py | 9 ---------
 2 files changed, 13 deletions(-)

diff --git a/emmet-core/emmet/core/settings.py b/emmet-core/emmet/core/settings.py
index 4f096f0f84..68a1497394 100644
--- a/emmet-core/emmet/core/settings.py
+++ b/emmet-core/emmet/core/settings.py
@@ -79,10 +79,6 @@ class EmmetSettings(BaseSettings):
         description="Maximum upward gradient in the last SCF for any VASP calculation",
     )
 
-    VASP_DEPRECATED_TAGS: List[str] = Field(
-        [], description="List of tags to manually deprecate in task validation"
-    )
-
     VASP_USE_STATICS: bool = Field(
         True,
         description="Use static calculations for structure and energy along with structure optimizations",
diff --git a/emmet-core/emmet/core/vasp/validation.py b/emmet-core/emmet/core/vasp/validation.py
index ab3d46d795..9beb59d9bd 100644
--- a/emmet-core/emmet/core/vasp/validation.py
+++ b/emmet-core/emmet/core/vasp/validation.py
@@ -13,7 +13,6 @@
 
 class DeprecationMessage(DocEnum):
     MANUAL = "M", "manual deprecation"
-    DEPRECATED_TAGS = "M001", "Deprecated tag"
     KPTS = "C001", "Too few KPoints"
     KSPACING = "C002", "KSpacing not high enough"
     ENCUT = "C002", "ENCUT too low"
@@ -57,7 +56,6 @@ def from_task_doc(
         input_sets: Dict[str, PyObject] = SETTINGS.VASP_DEFAULT_INPUT_SETS,
         LDAU_fields: List[str] = SETTINGS.VASP_CHECKED_LDAU_FIELDS,
         max_allowed_scf_gradient: float = SETTINGS.VASP_MAX_SCF_GRADIENT,
-        deprecated_tags: List[str] = SETTINGS.VASP_DEPRECATED_TAGS,
     ) -> "ValidationDoc":
         """
         Determines if a calculation is valid based on expected input parameters from a pymatgen inputset
@@ -168,13 +166,6 @@ def from_task_doc(
         if max_gradient > max_allowed_scf_gradient:
             reasons.append(DeprecationMessage.MAX_SCF)
 
-        # Check for Manual deprecations
-        if deprecated_tags is not None:
-            bad_tags = list(set(task_doc.tags).intersection(deprecated_tags))
-            if len(bad_tags) > 0:
-                reasons.append(DeprecationMessage.DEPRECATED_TAGS)
-                data["bad_tags"] = bad_tags
-
         doc = ValidationDoc(
             task_id=task_doc.task_id,
             calc_type=calc_type,

From 870dad427c2ebbf0c35323477ff1cfc7e827a880 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:01:41 -0700
Subject: [PATCH 09/26] Add oxidation states to thermo builder

---
 emmet-builders/emmet/builders/vasp/thermo.py | 26 ++++++++++++++++++--
 1 file changed, 24 insertions(+), 2 deletions(-)

diff --git a/emmet-builders/emmet/builders/vasp/thermo.py b/emmet-builders/emmet/builders/vasp/thermo.py
index 191a9955d6..b7f5df8def 100644
--- a/emmet-builders/emmet/builders/vasp/thermo.py
+++ b/emmet-builders/emmet/builders/vasp/thermo.py
@@ -25,6 +25,7 @@ def __init__(
         self,
         materials: Store,
         thermo: Store,
+        oxidation_states: Optional[Store] = None,
         query: Optional[Dict] = None,
         compatibility=None,
         **kwargs,
@@ -48,11 +49,17 @@ def __init__(
         self.compatibility = (
             compatibility
             if compatibility
-            else MaterialsProjectCompatibility("Advanced")
+            else MaterialsProject2020Compatibility("Advanced")
         )
+        self.oxidation_states = oxidation_states
         self._completed_tasks = set()
         self._entries_cache = defaultdict(list)
-        super().__init__(sources=[materials], targets=[thermo], **kwargs)
+
+        sources = [materials]
+        if oxidation_states is not None:
+            sources.append(oxidation_states)
+
+        super().__init__(sources=sources, targets=[thermo], **kwargs)
 
     def ensure_indexes(self):
         """
@@ -212,6 +219,18 @@ def get_entries(self, chemsys: str) -> List[Dict]:
             )
         )
 
+        # Get Oxidation state data for each material
+        oxi_states_data = {}
+        if self.oxidation_states:
+            material_ids = [t["material_id"] for t in materials_docs]
+            oxi_states_data = {
+                d["material_id"]: d["bond_valence"]["average_oxidation_states"]
+                for d in self.oxidation_states.query(
+                    properties=["material_id", "bond_valence.average_oxidation_states"],
+                    criteria={"material_id": {"$in": material_ids}, "successful": True},
+                )
+            }
+
         self.logger.debug(
             f"Got {len(materials_docs)} entries from DB for {len(query_chemsys)} sub-chemsys for {chemsys}"
         )
@@ -219,6 +238,9 @@ def get_entries(self, chemsys: str) -> List[Dict]:
         # Convert the entries into ComputedEntries and store
         for doc in materials_docs:
             for r_type, entry_dict in doc.get("entries", {}).items():
+                entry_dict["data"]["oxidation_states"] = oxi_states_data.get(
+                    entry_dict["entry_id"], {}
+                )
                 entry_dict["data"]["run_type"] = r_type
                 elsyms = sorted(set([el for el in entry_dict["composition"]]))
                 self._entries_cache["-".join(elsyms)].append(entry_dict)

From 08db4464c5f0445710466b1c1eae7e6c51847321 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:02:05 -0700
Subject: [PATCH 10/26] add prechunk to thermo and materials

---
 .../emmet/builders/vasp/materials.py          | 30 +++++++++++++++++++
 emmet-builders/emmet/builders/vasp/thermo.py  | 15 ++++++++++
 2 files changed, 45 insertions(+)

diff --git a/emmet-builders/emmet/builders/vasp/materials.py b/emmet-builders/emmet/builders/vasp/materials.py
index fe6f10608b..9608ae1935 100644
--- a/emmet-builders/emmet/builders/vasp/materials.py
+++ b/emmet-builders/emmet/builders/vasp/materials.py
@@ -91,6 +91,36 @@ def ensure_indexes(self):
             self.task_validation.ensure_index("task_id")
             self.task_validation.ensure_index("valid")
 
+    def prechunk(self, number_splits: int) -> Iterable[Dict]:
+        """Prechunk the materials builder for distributed computation"""
+        temp_query = dict(self.query)
+        temp_query["state"] = "successful"
+        if len(self.settings.BUILD_TAGS) > 0 and len(self.settings.EXCLUDED_TAGS) > 0:
+            temp_query["$and"] = [
+                {"tags": {"$in": self.settings.BUILD_TAGS}},
+                {"tags": {"$nin": self.settings.EXCLUDED_TAGS}},
+            ]
+        elif len(self.settings.BUILD_TAGS) > 0:
+            temp_query["tags"] = {"$in": self.settings.BUILD_TAGS}
+
+        self.logger.info("Finding tasks to process")
+        all_tasks = {
+            doc[self.tasks.key]
+            for doc in self.tasks.query(temp_query, [self.tasks.key])
+        }
+        processed_tasks = {
+            t_id
+            for d in self.materials.query({}, ["task_ids"])
+            for t_id in d.get("task_ids", [])
+        }
+        to_process_tasks = all_tasks - processed_tasks
+        to_process_forms = self.tasks.distinct(
+            "formula_pretty", {self.tasks.key: {"$in": list(to_process_tasks)}}
+        )
+
+        for formula_chunk in grouper(to_process_forms, number_splits):
+            yield {"formula_pretty": {"$in": list(formula_chunk)}}
+
     def get_items(self) -> Iterator[List[Dict]]:
         """
         Gets all items to process into materials documents.
diff --git a/emmet-builders/emmet/builders/vasp/thermo.py b/emmet-builders/emmet/builders/vasp/thermo.py
index b7f5df8def..e1f23f4538 100644
--- a/emmet-builders/emmet/builders/vasp/thermo.py
+++ b/emmet-builders/emmet/builders/vasp/thermo.py
@@ -74,6 +74,21 @@ def ensure_indexes(self):
         self.thermo.ensure_index("material_id")
         self.thermo.ensure_index("last_updated")
 
+    def prechunk(self, number_splits: int) -> Iterable[Dict]:
+        updated_chemsys = self.get_updated_chemsys()
+        new_chemsys = self.get_new_chemsys()
+
+        affected_chemsys = self.get_affected_chemsys(updated_chemsys | new_chemsys)
+
+        # Remove overlapping chemical systems
+        to_process_chemsys = set()
+        for chemsys in updated_chemsys | new_chemsys | affected_chemsys:
+            if chemsys not in to_process_chemsys:
+                to_process_chemsys |= chemsys_permutations(chemsys)
+
+        for chemsys_chunk in grouper(to_process_chemsys, number_splits):
+            yield {"chemsys": {"$in": list(chemsys_chunk)}}
+
     def get_items(self) -> Iterator[List[Dict]]:
         """
         Gets whole chemical systems of entries to process

From d0fc4351efe4ef6f886490b694c60e809626f645 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:02:23 -0700
Subject: [PATCH 11/26] jsanitize in process to not bog down main process

---
 emmet-builders/emmet/builders/vasp/materials.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/emmet-builders/emmet/builders/vasp/materials.py b/emmet-builders/emmet/builders/vasp/materials.py
index 9608ae1935..a186553491 100644
--- a/emmet-builders/emmet/builders/vasp/materials.py
+++ b/emmet-builders/emmet/builders/vasp/materials.py
@@ -243,7 +243,7 @@ def process_item(self, tasks: List[Dict]) -> List[Dict]:
                 )
         self.logger.debug(f"Produced {len(materials)} materials for {formula}")
 
-        return [mat.dict() for mat in materials]
+        return jsanitize([mat.dict() for mat in materials], allow_bson=True)
 
     def update_targets(self, items: List[List[Dict]]):
         """
@@ -265,7 +265,7 @@ def update_targets(self, items: List[List[Dict]]):
             self.logger.info(f"Updating {len(items)} materials")
             self.materials.remove_docs({self.materials.key: {"$in": material_ids}})
             self.materials.update(
-                docs=jsanitize(items, allow_bson=True),
+                docs=items,
                 key=["material_id"],
             )
         else:

From e74f55fe432c5b59831ed80724cae2b0d3d03ab9 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:02:44 -0700
Subject: [PATCH 12/26] remove vestige of sandbox

---
 emmet-builders/emmet/builders/vasp/materials.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/emmet-builders/emmet/builders/vasp/materials.py b/emmet-builders/emmet/builders/vasp/materials.py
index a186553491..00ef9e563d 100644
--- a/emmet-builders/emmet/builders/vasp/materials.py
+++ b/emmet-builders/emmet/builders/vasp/materials.py
@@ -84,7 +84,6 @@ def ensure_indexes(self):
         # Search index for materials
         self.materials.ensure_index("material_id")
         self.materials.ensure_index("last_updated")
-        self.materials.ensure_index("sandboxes")
         self.materials.ensure_index("task_ids")
 
         if self.task_validation:

From aad5eeb5b67653d7f80e64317de6836048e17780 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:03:06 -0700
Subject: [PATCH 13/26] clean up

---
 .../emmet/builders/vasp/materials.py          | 20 +++--------
 .../emmet/builders/vasp/task_validator.py     |  4 +--
 emmet-builders/emmet/builders/vasp/thermo.py  | 33 +++++++------------
 3 files changed, 19 insertions(+), 38 deletions(-)

diff --git a/emmet-builders/emmet/builders/vasp/materials.py b/emmet-builders/emmet/builders/vasp/materials.py
index 00ef9e563d..c9f76560ff 100644
--- a/emmet-builders/emmet/builders/vasp/materials.py
+++ b/emmet-builders/emmet/builders/vasp/materials.py
@@ -1,21 +1,13 @@
 from datetime import datetime
 from itertools import chain
-from operator import itemgetter
-from typing import Dict, Iterator, List, Optional
+from typing import Dict, Iterable, Iterator, List, Optional
 
 from maggma.builders import Builder
 from maggma.stores import Store
-from pymatgen.core import Structure
-from pymatgen.analysis.structure_analyzer import oxide_type
-from pymatgen.analysis.structure_matcher import ElementComparator, StructureMatcher
+from maggma.utils import grouper
 
-from emmet.builders.utils import maximal_spanning_non_intersecting_subsets
-
-# from emmet.core import SETTINGS
-from emmet.builders import SETTINGS
 from emmet.builders.settings import EmmetBuildSettings
 from emmet.core.utils import group_structures, jsanitize
-from emmet.core.vasp.calc_types import TaskType
 from emmet.core.vasp.material import MaterialsDoc
 from emmet.core.vasp.task import TaskDocument
 
@@ -33,11 +25,8 @@ class MaterialsBuilder(Builder):
         1.) Find all documents with the same formula
         2.) Select only task documents for the task_types we can select properties from
         3.) Aggregate task documents based on strucutre similarity
-        4.) Convert task docs to property docs with metadata for selection and aggregation
-        5.) Select the best property doc for each property
-        6.) Build material document from best property docs
-        7.) Post-process material document
-        8.) Validate material document
+        4.) Create a MaterialDoc from the group of task documents
+        5.) Validate material document
 
     """
 
@@ -56,6 +45,7 @@ def __init__(
             materials: Store of materials documents to generate
             task_validation: Store for storing task validation results
             query: dictionary to limit tasks to be analyzed
+            settings: EmmetSettings to use in the build process
         """
 
         self.tasks = tasks
diff --git a/emmet-builders/emmet/builders/vasp/task_validator.py b/emmet-builders/emmet/builders/vasp/task_validator.py
index 82843f23b4..1f8d368d8f 100644
--- a/emmet-builders/emmet/builders/vasp/task_validator.py
+++ b/emmet-builders/emmet/builders/vasp/task_validator.py
@@ -1,4 +1,4 @@
-from typing import Dict, List, Union, Optional
+from typing import Dict, List, Optional, Union
 
 import numpy as np
 from maggma.builders import MapBuilder
@@ -6,10 +6,10 @@
 from pymatgen.core import Structure
 
 from emmet.builders import SETTINGS
+from emmet.builders.settings import EmmetBuildSettings
 from emmet.core.vasp.calc_types import run_type, task_type
 from emmet.core.vasp.task import TaskDocument
 from emmet.core.vasp.validation import DeprecationMessage, ValidationDoc
-from emmet.builders.settings import EmmetBuildSettings
 
 
 class TaskValidator(MapBuilder):
diff --git a/emmet-builders/emmet/builders/vasp/thermo.py b/emmet-builders/emmet/builders/vasp/thermo.py
index e1f23f4538..cba5a262c7 100644
--- a/emmet-builders/emmet/builders/vasp/thermo.py
+++ b/emmet-builders/emmet/builders/vasp/thermo.py
@@ -1,23 +1,17 @@
 from collections import defaultdict
-from datetime import datetime
 from itertools import chain
-from typing import Dict, Iterator, List, Optional, Set, Tuple
+from typing import Dict, Iterable, Iterator, List, Optional, Set, Tuple
 
 from maggma.core import Builder, Store
+from maggma.utils import grouper
 from monty.json import MontyDecoder
-from pymatgen.core import Structure
 from pymatgen.analysis.phase_diagram import PhaseDiagramError
-from pymatgen.analysis.structure_analyzer import oxide_type
-from pymatgen.entries.compatibility import MaterialsProjectCompatibility
+from pymatgen.entries.compatibility import MaterialsProject2020Compatibility
 from pymatgen.entries.computed_entries import ComputedEntry, ComputedStructureEntry
 
-from emmet.core.utils import jsanitize
-from emmet.builders.utils import (
-    chemsys_permutations,
-    maximal_spanning_non_intersecting_subsets,
-)
+from emmet.builders.utils import chemsys_permutations
 from emmet.core.thermo import ThermoDoc
-from emmet.core.vasp.calc_types import run_type
+from emmet.core.utils import jsanitize
 
 
 class Thermo(Builder):
@@ -121,13 +115,12 @@ def get_items(self) -> Iterator[List[Dict]]:
             entries = self.get_entries(chemsys)
             yield entries
 
-    def process_item(self, item: Tuple[List[str], List[ComputedEntry]]):
+    def process_item(self, item: List[Dict]):
 
-        entries = item
-        if len(entries) == 0:
+        if len(item) == 0:
             return []
 
-        entries = [ComputedStructureEntry.from_dict(entry) for entry in entries]
+        entries = [ComputedStructureEntry.from_dict(entry) for entry in item]
         # determine chemsys
         elements = sorted(
             set([el.symbol for e in entries for el in e.composition.elements])
@@ -229,9 +222,7 @@ def get_entries(self, chemsys: str) -> List[Dict]:
         new_q["chemsys"] = {"$in": list(query_chemsys)}
         new_q["deprecated"] = False
         materials_docs = list(
-            self.materials.query(
-                criteria=new_q, properties=[self.materials.key, "entries"]
-            )
+            self.materials.query(criteria=new_q, properties=["material_id", "entries"])
         )
 
         # Get Oxidation state data for each material
@@ -268,7 +259,7 @@ def get_entries(self, chemsys: str) -> List[Dict]:
     def get_updated_chemsys(
         self,
     ) -> Set:
-        """ Gets updated chemical system as defined by the updating of an existing material """
+        """Gets updated chemical system as defined by the updating of an existing material"""
 
         updated_mats = self.thermo.newer_in(self.materials, criteria=self.query)
         updated_chemsys = set(
@@ -281,7 +272,7 @@ def get_updated_chemsys(
         return updated_chemsys
 
     def get_new_chemsys(self) -> Set:
-        """ Gets newer chemical system as defined by introduction of a new material """
+        """Gets newer chemical system as defined by introduction of a new material"""
 
         # All materials that are not present in the thermo collection
         thermo_mat_ids = self.thermo.distinct(self.thermo.key)
@@ -294,7 +285,7 @@ def get_new_chemsys(self) -> Set:
         return new_mat_chemsys
 
     def get_affected_chemsys(self, chemical_systems: Set) -> Set:
-        """ Gets chemical systems affected by changes in the supplied chemical systems """
+        """Gets chemical systems affected by changes in the supplied chemical systems"""
         # First get all chemsys with any of the elements we've marked
         affected_chemsys = set()
         affected_els = list({el for c in chemical_systems for el in c.split("-")})

From 0791b3be65a6742997e591b9b32b3effae4d3683 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:12:20 -0700
Subject: [PATCH 14/26] reraise exception if failure

---
 emmet-core/emmet/core/oxidation_states.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/emmet-core/emmet/core/oxidation_states.py b/emmet-core/emmet/core/oxidation_states.py
index 41ecfc1a50..46f120f96c 100644
--- a/emmet-core/emmet/core/oxidation_states.py
+++ b/emmet-core/emmet/core/oxidation_states.py
@@ -85,3 +85,4 @@ def from_structure(cls, structure: Structure):
 
             except Exception as e:
                 logging.error("Oxidation state guess failed with: {}".format(e))
+                raise e

From 801f920776ac2895eb2b0d2c38d0269f69ce693a Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:12:28 -0700
Subject: [PATCH 15/26] add oxidation state builder

---
 .../builders/materials/oxidation_states.py    | 55 +++++++++++++++++++
 1 file changed, 55 insertions(+)
 create mode 100644 emmet-builders/emmet/builders/materials/oxidation_states.py

diff --git a/emmet-builders/emmet/builders/materials/oxidation_states.py b/emmet-builders/emmet/builders/materials/oxidation_states.py
new file mode 100644
index 0000000000..46a726e326
--- /dev/null
+++ b/emmet-builders/emmet/builders/materials/oxidation_states.py
@@ -0,0 +1,55 @@
+from typing import Dict, List, Optional
+
+from maggma.builders.map_builder import MapBuilder
+from maggma.core import Store
+from pymatgen.core import Structure
+
+from pymatgen.core import __version__ as pymatgen_version
+
+
+from emmet.core.oxidation_states import OxidationStateDoc
+
+
+class OxidationStates(MapBuilder):
+    def __init__(
+        self,
+        materials: Store,
+        oxidation_states: Store,
+        **kwargs,
+    ):
+        """
+        Creates Oxidation State documents from materials
+
+        Args:
+            materials: Store of materials docs
+            oxidation_states: Store to update with oxidation state document
+            query : query on materials to limit search
+        """
+        self.materials = materials
+        self.oxidation_states = oxidation_states
+        self.kwargs = kwargs
+
+        # Enforce that we key on material_id
+        self.materials.key = "material_id"
+        self.oxidation_states.key = "material_id"
+        super().__init__(
+            source=materials,
+            target=oxidation_states,
+            projection=["structure"],
+            **kwargs,
+        )
+
+    def unary_function(self, item):
+
+        structure = Structure.from_dict(item["structure"])
+        oxi_doc = OxidationStateDoc.from_structure(structure)
+        doc = oxi_doc.dict()
+
+        doc.update(
+            {
+                "pymatgen_version": pymatgen_version,
+                "successful": True,
+            }
+        )
+
+        return doc

From 9e47131f30c1e3d750ebacc062b8d619e60ced75 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:32:43 -0700
Subject: [PATCH 16/26] compatability will select run_type

---
 emmet-builders/emmet/builders/vasp/thermo.py | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/emmet-builders/emmet/builders/vasp/thermo.py b/emmet-builders/emmet/builders/vasp/thermo.py
index cba5a262c7..51ecfb3964 100644
--- a/emmet-builders/emmet/builders/vasp/thermo.py
+++ b/emmet-builders/emmet/builders/vasp/thermo.py
@@ -134,13 +134,7 @@ def process_item(self, item: List[Dict]):
         for entry in entries:
             material_entries[entry.entry_id][entry.data["run_type"]] = entry
 
-        # TODO: How to make this general and controllable via SETTINGS?
-        for material_id in material_entries:
-            if "GGA+U" in material_entries[material_id]:
-                pd_entries.append(material_entries[material_id]["GGA+U"])
-            elif "GGA" in material_entries[material_id]:
-                pd_entries.append(material_entries[material_id]["GGA"])
-        pd_entries = self.compatibility.process_entries(pd_entries)
+        pd_entries = self.compatibility.process_entries(entries)
         self.logger.debug(f"{len(pd_entries)} remain in {chemsys} after filtering")
 
         try:

From 00e4cce2233325033931fd2d62c2f19c1d273805 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:33:01 -0700
Subject: [PATCH 17/26] clean up

---
 emmet-builders/emmet/builders/vasp/thermo.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/emmet-builders/emmet/builders/vasp/thermo.py b/emmet-builders/emmet/builders/vasp/thermo.py
index 51ecfb3964..8b87da3835 100644
--- a/emmet-builders/emmet/builders/vasp/thermo.py
+++ b/emmet-builders/emmet/builders/vasp/thermo.py
@@ -158,7 +158,7 @@ def process_item(self, item: List[Dict]):
             )
             return []
 
-        return [d.dict() for d in docs]
+        return jsanitize([d.dict() for d in docs], allow_bson=True)
 
     def update_targets(self, items):
         """
@@ -169,9 +169,9 @@ def update_targets(self, items):
         # flatten out lists
         items = list(filter(None, chain.from_iterable(items)))
         # Check if already updated this run
-        items = [i for i in items if i[self.thermo.key] not in self._completed_tasks]
+        items = [i for i in items if i["material_id"] not in self._completed_tasks]
 
-        self._completed_tasks |= {i[self.thermo.key] for i in items}
+        self._completed_tasks |= {i["material_id"] for i in items}
 
         for item in items:
             if isinstance(item["last_updated"], dict):
@@ -181,7 +181,7 @@ def update_targets(self, items):
 
         if len(items) > 0:
             self.logger.info(f"Updating {len(items)} thermo documents")
-            self.thermo.update(docs=jsanitize(items), key=[self.thermo.key])
+            self.thermo.update(docs=items, key=["material_id"])
         else:
             self.logger.info("No items to update")
 

From d7cab8aab595325df9778be1934d87c93e760732 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 15:46:24 -0700
Subject: [PATCH 18/26] update builder names

---
 emmet-builders/emmet/builders/materials/oxidation_states.py | 3 +--
 emmet-builders/emmet/builders/vasp/thermo.py                | 2 +-
 2 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/emmet-builders/emmet/builders/materials/oxidation_states.py b/emmet-builders/emmet/builders/materials/oxidation_states.py
index 46a726e326..b250fdec3a 100644
--- a/emmet-builders/emmet/builders/materials/oxidation_states.py
+++ b/emmet-builders/emmet/builders/materials/oxidation_states.py
@@ -10,7 +10,7 @@
 from emmet.core.oxidation_states import OxidationStateDoc
 
 
-class OxidationStates(MapBuilder):
+class OxidationStatesBuilder(MapBuilder):
     def __init__(
         self,
         materials: Store,
@@ -40,7 +40,6 @@ def __init__(
         )
 
     def unary_function(self, item):
-
         structure = Structure.from_dict(item["structure"])
         oxi_doc = OxidationStateDoc.from_structure(structure)
         doc = oxi_doc.dict()
diff --git a/emmet-builders/emmet/builders/vasp/thermo.py b/emmet-builders/emmet/builders/vasp/thermo.py
index 8b87da3835..c237ff2b2f 100644
--- a/emmet-builders/emmet/builders/vasp/thermo.py
+++ b/emmet-builders/emmet/builders/vasp/thermo.py
@@ -14,7 +14,7 @@
 from emmet.core.utils import jsanitize
 
 
-class Thermo(Builder):
+class ThermoBuilder(Builder):
     def __init__(
         self,
         materials: Store,

From b119add3d8e6772951d6557265e08cc5eed1dec6 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 16:06:54 -0700
Subject: [PATCH 19/26] fix misc bugs

---
 emmet-builders/emmet/builders/vasp/task_validator.py | 4 ++--
 emmet-core/emmet/core/vasp/validation.py             | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/emmet-builders/emmet/builders/vasp/task_validator.py b/emmet-builders/emmet/builders/vasp/task_validator.py
index 1f8d368d8f..66226ac0e1 100644
--- a/emmet-builders/emmet/builders/vasp/task_validator.py
+++ b/emmet-builders/emmet/builders/vasp/task_validator.py
@@ -40,8 +40,8 @@ def __init__(
             projection=[
                 "orig_inputs",
                 "output.structure",
-                "input.parameters",
-                "calcs_reversed.output.ionic_steps.e_fr_energy",
+                "output.bandgap" "input.parameters",
+                "calcs_reversed.output.ionic_steps.electronic_steps.e_fr_energy",
                 "tags",
             ],
             query=query,
diff --git a/emmet-core/emmet/core/vasp/validation.py b/emmet-core/emmet/core/vasp/validation.py
index 9beb59d9bd..9846c3624b 100644
--- a/emmet-core/emmet/core/vasp/validation.py
+++ b/emmet-core/emmet/core/vasp/validation.py
@@ -116,7 +116,7 @@ def from_task_doc(
                         )
 
             # warn, but don't invalidate if wrong ISMEAR
-            if inputs.get("incar", {}).get("ISMEAR") > 0 and bandgap > 0:
+            if inputs.get("incar", {}).get("ISMEAR", 1) > 0 and bandgap > 0:
                 warnings.append("Inappropriate smearing settings")
 
             # Checking ENCUT

From 30ce56a842f8a83bcec34b0446d22fbab230ed54 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Thu, 27 May 2021 16:07:07 -0700
Subject: [PATCH 20/26] formatting bug

---
 emmet-builders/emmet/builders/vasp/task_validator.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/emmet-builders/emmet/builders/vasp/task_validator.py b/emmet-builders/emmet/builders/vasp/task_validator.py
index 66226ac0e1..08610f2993 100644
--- a/emmet-builders/emmet/builders/vasp/task_validator.py
+++ b/emmet-builders/emmet/builders/vasp/task_validator.py
@@ -40,7 +40,8 @@ def __init__(
             projection=[
                 "orig_inputs",
                 "output.structure",
-                "output.bandgap" "input.parameters",
+                "output.bandgap",
+                "input.parameters",
                 "calcs_reversed.output.ionic_steps.electronic_steps.e_fr_energy",
                 "tags",
             ],

From ddf6fa1fde4bbb1508468bc8e204f535cbf30d47 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Fri, 28 May 2021 13:47:04 -0700
Subject: [PATCH 21/26] mark when ismear differs

---
 emmet-core/emmet/core/vasp/validation.py | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/emmet-core/emmet/core/vasp/validation.py b/emmet-core/emmet/core/vasp/validation.py
index 9846c3624b..6dcfdc35bd 100644
--- a/emmet-core/emmet/core/vasp/validation.py
+++ b/emmet-core/emmet/core/vasp/validation.py
@@ -116,8 +116,13 @@ def from_task_doc(
                         )
 
             # warn, but don't invalidate if wrong ISMEAR
-            if inputs.get("incar", {}).get("ISMEAR", 1) > 0 and bandgap > 0:
-                warnings.append("Inappropriate smearing settings")
+            valid_ismear = valid_input_set.incar.get("ISMEAR", 1)
+            curr_ismear = inputs.get("incar", {}).get("ISMEAR", 1)
+            if curr_ismear != valid_ismear:
+                warnings.append(
+                    f"Inappropriate smearing settings. Set to {curr_ismear},"
+                    " but should be {valid_ismear}"
+                )
 
             # Checking ENCUT
             encut = inputs.get("incar", {}).get("ENCUT")

From 8bc1999d3b947b8716d19af46e626be9ce3fafab Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Fri, 28 May 2021 13:48:08 -0700
Subject: [PATCH 22/26] add unrecognized task_type

---
 .../emmet/core/vasp/calc_types/enums.py       | 60 +++++++++++++++++++
 .../emmet/core/vasp/calc_types/generate.py    | 15 +----
 2 files changed, 62 insertions(+), 13 deletions(-)

diff --git a/emmet-core/emmet/core/vasp/calc_types/enums.py b/emmet-core/emmet/core/vasp/calc_types/enums.py
index 35bb01b22d..5c0241e80f 100644
--- a/emmet-core/emmet/core/vasp/calc_types/enums.py
+++ b/emmet-core/emmet/core/vasp/calc_types/enums.py
@@ -99,6 +99,7 @@ class CalcType(ValueEnum):
     AM05_Static = "AM05 Static"
     AM05_Structure_Optimization = "AM05 Structure Optimization"
     AM05_Deformation = "AM05 Deformation"
+    AM05_Unrecognized = "AM05 Unrecognized"
     GGA_NSCF_Line = "GGA NSCF Line"
     GGA_NSCF_Uniform = "GGA NSCF Uniform"
     GGA_Dielectric = "GGA Dielectric"
@@ -109,6 +110,7 @@ class CalcType(ValueEnum):
     GGA_Static = "GGA Static"
     GGA_Structure_Optimization = "GGA Structure Optimization"
     GGA_Deformation = "GGA Deformation"
+    GGA_Unrecognized = "GGA Unrecognized"
     PBE_NSCF_Line = "PBE NSCF Line"
     PBE_NSCF_Uniform = "PBE NSCF Uniform"
     PBE_Dielectric = "PBE Dielectric"
@@ -119,6 +121,7 @@ class CalcType(ValueEnum):
     PBE_Static = "PBE Static"
     PBE_Structure_Optimization = "PBE Structure Optimization"
     PBE_Deformation = "PBE Deformation"
+    PBE_Unrecognized = "PBE Unrecognized"
     PBESol_NSCF_Line = "PBESol NSCF Line"
     PBESol_NSCF_Uniform = "PBESol NSCF Uniform"
     PBESol_Dielectric = "PBESol Dielectric"
@@ -129,6 +132,7 @@ class CalcType(ValueEnum):
     PBESol_Static = "PBESol Static"
     PBESol_Structure_Optimization = "PBESol Structure Optimization"
     PBESol_Deformation = "PBESol Deformation"
+    PBESol_Unrecognized = "PBESol Unrecognized"
     RevPBE_PADE_NSCF_Line = "RevPBE+PADE NSCF Line"
     RevPBE_PADE_NSCF_Uniform = "RevPBE+PADE NSCF Uniform"
     RevPBE_PADE_Dielectric = "RevPBE+PADE Dielectric"
@@ -139,6 +143,7 @@ class CalcType(ValueEnum):
     RevPBE_PADE_Static = "RevPBE+PADE Static"
     RevPBE_PADE_Structure_Optimization = "RevPBE+PADE Structure Optimization"
     RevPBE_PADE_Deformation = "RevPBE+PADE Deformation"
+    RevPBE_PADE_Unrecognized = "RevPBE+PADE Unrecognized"
     optB86b_NSCF_Line = "optB86b NSCF Line"
     optB86b_NSCF_Uniform = "optB86b NSCF Uniform"
     optB86b_Dielectric = "optB86b Dielectric"
@@ -149,6 +154,7 @@ class CalcType(ValueEnum):
     optB86b_Static = "optB86b Static"
     optB86b_Structure_Optimization = "optB86b Structure Optimization"
     optB86b_Deformation = "optB86b Deformation"
+    optB86b_Unrecognized = "optB86b Unrecognized"
     optB88_NSCF_Line = "optB88 NSCF Line"
     optB88_NSCF_Uniform = "optB88 NSCF Uniform"
     optB88_Dielectric = "optB88 Dielectric"
@@ -159,6 +165,7 @@ class CalcType(ValueEnum):
     optB88_Static = "optB88 Static"
     optB88_Structure_Optimization = "optB88 Structure Optimization"
     optB88_Deformation = "optB88 Deformation"
+    optB88_Unrecognized = "optB88 Unrecognized"
     optPBE_NSCF_Line = "optPBE NSCF Line"
     optPBE_NSCF_Uniform = "optPBE NSCF Uniform"
     optPBE_Dielectric = "optPBE Dielectric"
@@ -169,6 +176,7 @@ class CalcType(ValueEnum):
     optPBE_Static = "optPBE Static"
     optPBE_Structure_Optimization = "optPBE Structure Optimization"
     optPBE_Deformation = "optPBE Deformation"
+    optPBE_Unrecognized = "optPBE Unrecognized"
     revPBE_NSCF_Line = "revPBE NSCF Line"
     revPBE_NSCF_Uniform = "revPBE NSCF Uniform"
     revPBE_Dielectric = "revPBE Dielectric"
@@ -179,6 +187,7 @@ class CalcType(ValueEnum):
     revPBE_Static = "revPBE Static"
     revPBE_Structure_Optimization = "revPBE Structure Optimization"
     revPBE_Deformation = "revPBE Deformation"
+    revPBE_Unrecognized = "revPBE Unrecognized"
     B3LYP_NSCF_Line = "B3LYP NSCF Line"
     B3LYP_NSCF_Uniform = "B3LYP NSCF Uniform"
     B3LYP_Dielectric = "B3LYP Dielectric"
@@ -189,6 +198,7 @@ class CalcType(ValueEnum):
     B3LYP_Static = "B3LYP Static"
     B3LYP_Structure_Optimization = "B3LYP Structure Optimization"
     B3LYP_Deformation = "B3LYP Deformation"
+    B3LYP_Unrecognized = "B3LYP Unrecognized"
     HF_NSCF_Line = "HF NSCF Line"
     HF_NSCF_Uniform = "HF NSCF Uniform"
     HF_Dielectric = "HF Dielectric"
@@ -199,6 +209,7 @@ class CalcType(ValueEnum):
     HF_Static = "HF Static"
     HF_Structure_Optimization = "HF Structure Optimization"
     HF_Deformation = "HF Deformation"
+    HF_Unrecognized = "HF Unrecognized"
     HSE03_NSCF_Line = "HSE03 NSCF Line"
     HSE03_NSCF_Uniform = "HSE03 NSCF Uniform"
     HSE03_Dielectric = "HSE03 Dielectric"
@@ -209,6 +220,7 @@ class CalcType(ValueEnum):
     HSE03_Static = "HSE03 Static"
     HSE03_Structure_Optimization = "HSE03 Structure Optimization"
     HSE03_Deformation = "HSE03 Deformation"
+    HSE03_Unrecognized = "HSE03 Unrecognized"
     HSE06_NSCF_Line = "HSE06 NSCF Line"
     HSE06_NSCF_Uniform = "HSE06 NSCF Uniform"
     HSE06_Dielectric = "HSE06 Dielectric"
@@ -219,6 +231,7 @@ class CalcType(ValueEnum):
     HSE06_Static = "HSE06 Static"
     HSE06_Structure_Optimization = "HSE06 Structure Optimization"
     HSE06_Deformation = "HSE06 Deformation"
+    HSE06_Unrecognized = "HSE06 Unrecognized"
     PB0_NSCF_Line = "PB0 NSCF Line"
     PB0_NSCF_Uniform = "PB0 NSCF Uniform"
     PB0_Dielectric = "PB0 Dielectric"
@@ -229,6 +242,7 @@ class CalcType(ValueEnum):
     PB0_Static = "PB0 Static"
     PB0_Structure_Optimization = "PB0 Structure Optimization"
     PB0_Deformation = "PB0 Deformation"
+    PB0_Unrecognized = "PB0 Unrecognized"
     M06L_NSCF_Line = "M06L NSCF Line"
     M06L_NSCF_Uniform = "M06L NSCF Uniform"
     M06L_Dielectric = "M06L Dielectric"
@@ -239,6 +253,7 @@ class CalcType(ValueEnum):
     M06L_Static = "M06L Static"
     M06L_Structure_Optimization = "M06L Structure Optimization"
     M06L_Deformation = "M06L Deformation"
+    M06L_Unrecognized = "M06L Unrecognized"
     MBJL_NSCF_Line = "MBJL NSCF Line"
     MBJL_NSCF_Uniform = "MBJL NSCF Uniform"
     MBJL_Dielectric = "MBJL Dielectric"
@@ -249,6 +264,7 @@ class CalcType(ValueEnum):
     MBJL_Static = "MBJL Static"
     MBJL_Structure_Optimization = "MBJL Structure Optimization"
     MBJL_Deformation = "MBJL Deformation"
+    MBJL_Unrecognized = "MBJL Unrecognized"
     MS0_NSCF_Line = "MS0 NSCF Line"
     MS0_NSCF_Uniform = "MS0 NSCF Uniform"
     MS0_Dielectric = "MS0 Dielectric"
@@ -259,6 +275,7 @@ class CalcType(ValueEnum):
     MS0_Static = "MS0 Static"
     MS0_Structure_Optimization = "MS0 Structure Optimization"
     MS0_Deformation = "MS0 Deformation"
+    MS0_Unrecognized = "MS0 Unrecognized"
     MS1_NSCF_Line = "MS1 NSCF Line"
     MS1_NSCF_Uniform = "MS1 NSCF Uniform"
     MS1_Dielectric = "MS1 Dielectric"
@@ -269,6 +286,7 @@ class CalcType(ValueEnum):
     MS1_Static = "MS1 Static"
     MS1_Structure_Optimization = "MS1 Structure Optimization"
     MS1_Deformation = "MS1 Deformation"
+    MS1_Unrecognized = "MS1 Unrecognized"
     MS2_NSCF_Line = "MS2 NSCF Line"
     MS2_NSCF_Uniform = "MS2 NSCF Uniform"
     MS2_Dielectric = "MS2 Dielectric"
@@ -279,6 +297,7 @@ class CalcType(ValueEnum):
     MS2_Static = "MS2 Static"
     MS2_Structure_Optimization = "MS2 Structure Optimization"
     MS2_Deformation = "MS2 Deformation"
+    MS2_Unrecognized = "MS2 Unrecognized"
     RTPSS_NSCF_Line = "RTPSS NSCF Line"
     RTPSS_NSCF_Uniform = "RTPSS NSCF Uniform"
     RTPSS_Dielectric = "RTPSS Dielectric"
@@ -289,6 +308,7 @@ class CalcType(ValueEnum):
     RTPSS_Static = "RTPSS Static"
     RTPSS_Structure_Optimization = "RTPSS Structure Optimization"
     RTPSS_Deformation = "RTPSS Deformation"
+    RTPSS_Unrecognized = "RTPSS Unrecognized"
     SCAN_NSCF_Line = "SCAN NSCF Line"
     SCAN_NSCF_Uniform = "SCAN NSCF Uniform"
     SCAN_Dielectric = "SCAN Dielectric"
@@ -299,6 +319,7 @@ class CalcType(ValueEnum):
     SCAN_Static = "SCAN Static"
     SCAN_Structure_Optimization = "SCAN Structure Optimization"
     SCAN_Deformation = "SCAN Deformation"
+    SCAN_Unrecognized = "SCAN Unrecognized"
     TPSS_NSCF_Line = "TPSS NSCF Line"
     TPSS_NSCF_Uniform = "TPSS NSCF Uniform"
     TPSS_Dielectric = "TPSS Dielectric"
@@ -309,6 +330,7 @@ class CalcType(ValueEnum):
     TPSS_Static = "TPSS Static"
     TPSS_Structure_Optimization = "TPSS Structure Optimization"
     TPSS_Deformation = "TPSS Deformation"
+    TPSS_Unrecognized = "TPSS Unrecognized"
     SCAN_rVV10_NSCF_Line = "SCAN-rVV10 NSCF Line"
     SCAN_rVV10_NSCF_Uniform = "SCAN-rVV10 NSCF Uniform"
     SCAN_rVV10_Dielectric = "SCAN-rVV10 Dielectric"
@@ -319,6 +341,7 @@ class CalcType(ValueEnum):
     SCAN_rVV10_Static = "SCAN-rVV10 Static"
     SCAN_rVV10_Structure_Optimization = "SCAN-rVV10 Structure Optimization"
     SCAN_rVV10_Deformation = "SCAN-rVV10 Deformation"
+    SCAN_rVV10_Unrecognized = "SCAN-rVV10 Unrecognized"
     optB86b_vdW_NSCF_Line = "optB86b-vdW NSCF Line"
     optB86b_vdW_NSCF_Uniform = "optB86b-vdW NSCF Uniform"
     optB86b_vdW_Dielectric = "optB86b-vdW Dielectric"
@@ -329,6 +352,7 @@ class CalcType(ValueEnum):
     optB86b_vdW_Static = "optB86b-vdW Static"
     optB86b_vdW_Structure_Optimization = "optB86b-vdW Structure Optimization"
     optB86b_vdW_Deformation = "optB86b-vdW Deformation"
+    optB86b_vdW_Unrecognized = "optB86b-vdW Unrecognized"
     optB88_vdW_NSCF_Line = "optB88-vdW NSCF Line"
     optB88_vdW_NSCF_Uniform = "optB88-vdW NSCF Uniform"
     optB88_vdW_Dielectric = "optB88-vdW Dielectric"
@@ -339,6 +363,7 @@ class CalcType(ValueEnum):
     optB88_vdW_Static = "optB88-vdW Static"
     optB88_vdW_Structure_Optimization = "optB88-vdW Structure Optimization"
     optB88_vdW_Deformation = "optB88-vdW Deformation"
+    optB88_vdW_Unrecognized = "optB88-vdW Unrecognized"
     optPBE_vdW_NSCF_Line = "optPBE-vdW NSCF Line"
     optPBE_vdW_NSCF_Uniform = "optPBE-vdW NSCF Uniform"
     optPBE_vdW_Dielectric = "optPBE-vdW Dielectric"
@@ -349,6 +374,7 @@ class CalcType(ValueEnum):
     optPBE_vdW_Static = "optPBE-vdW Static"
     optPBE_vdW_Structure_Optimization = "optPBE-vdW Structure Optimization"
     optPBE_vdW_Deformation = "optPBE-vdW Deformation"
+    optPBE_vdW_Unrecognized = "optPBE-vdW Unrecognized"
     rev_vdW_DF2_NSCF_Line = "rev-vdW-DF2 NSCF Line"
     rev_vdW_DF2_NSCF_Uniform = "rev-vdW-DF2 NSCF Uniform"
     rev_vdW_DF2_Dielectric = "rev-vdW-DF2 Dielectric"
@@ -359,6 +385,7 @@ class CalcType(ValueEnum):
     rev_vdW_DF2_Static = "rev-vdW-DF2 Static"
     rev_vdW_DF2_Structure_Optimization = "rev-vdW-DF2 Structure Optimization"
     rev_vdW_DF2_Deformation = "rev-vdW-DF2 Deformation"
+    rev_vdW_DF2_Unrecognized = "rev-vdW-DF2 Unrecognized"
     revPBE_vdW_NSCF_Line = "revPBE-vdW NSCF Line"
     revPBE_vdW_NSCF_Uniform = "revPBE-vdW NSCF Uniform"
     revPBE_vdW_Dielectric = "revPBE-vdW Dielectric"
@@ -369,6 +396,7 @@ class CalcType(ValueEnum):
     revPBE_vdW_Static = "revPBE-vdW Static"
     revPBE_vdW_Structure_Optimization = "revPBE-vdW Structure Optimization"
     revPBE_vdW_Deformation = "revPBE-vdW Deformation"
+    revPBE_vdW_Unrecognized = "revPBE-vdW Unrecognized"
     vdW_DF2_NSCF_Line = "vdW-DF2 NSCF Line"
     vdW_DF2_NSCF_Uniform = "vdW-DF2 NSCF Uniform"
     vdW_DF2_Dielectric = "vdW-DF2 Dielectric"
@@ -379,6 +407,7 @@ class CalcType(ValueEnum):
     vdW_DF2_Static = "vdW-DF2 Static"
     vdW_DF2_Structure_Optimization = "vdW-DF2 Structure Optimization"
     vdW_DF2_Deformation = "vdW-DF2 Deformation"
+    vdW_DF2_Unrecognized = "vdW-DF2 Unrecognized"
     AM05_U_NSCF_Line = "AM05+U NSCF Line"
     AM05_U_NSCF_Uniform = "AM05+U NSCF Uniform"
     AM05_U_Dielectric = "AM05+U Dielectric"
@@ -389,6 +418,7 @@ class CalcType(ValueEnum):
     AM05_U_Static = "AM05+U Static"
     AM05_U_Structure_Optimization = "AM05+U Structure Optimization"
     AM05_U_Deformation = "AM05+U Deformation"
+    AM05_U_Unrecognized = "AM05+U Unrecognized"
     GGA_U_NSCF_Line = "GGA+U NSCF Line"
     GGA_U_NSCF_Uniform = "GGA+U NSCF Uniform"
     GGA_U_Dielectric = "GGA+U Dielectric"
@@ -399,6 +429,7 @@ class CalcType(ValueEnum):
     GGA_U_Static = "GGA+U Static"
     GGA_U_Structure_Optimization = "GGA+U Structure Optimization"
     GGA_U_Deformation = "GGA+U Deformation"
+    GGA_U_Unrecognized = "GGA+U Unrecognized"
     PBE_U_NSCF_Line = "PBE+U NSCF Line"
     PBE_U_NSCF_Uniform = "PBE+U NSCF Uniform"
     PBE_U_Dielectric = "PBE+U Dielectric"
@@ -409,6 +440,7 @@ class CalcType(ValueEnum):
     PBE_U_Static = "PBE+U Static"
     PBE_U_Structure_Optimization = "PBE+U Structure Optimization"
     PBE_U_Deformation = "PBE+U Deformation"
+    PBE_U_Unrecognized = "PBE+U Unrecognized"
     PBESol_U_NSCF_Line = "PBESol+U NSCF Line"
     PBESol_U_NSCF_Uniform = "PBESol+U NSCF Uniform"
     PBESol_U_Dielectric = "PBESol+U Dielectric"
@@ -419,6 +451,7 @@ class CalcType(ValueEnum):
     PBESol_U_Static = "PBESol+U Static"
     PBESol_U_Structure_Optimization = "PBESol+U Structure Optimization"
     PBESol_U_Deformation = "PBESol+U Deformation"
+    PBESol_U_Unrecognized = "PBESol+U Unrecognized"
     RevPBE_PADE_U_NSCF_Line = "RevPBE+PADE+U NSCF Line"
     RevPBE_PADE_U_NSCF_Uniform = "RevPBE+PADE+U NSCF Uniform"
     RevPBE_PADE_U_Dielectric = "RevPBE+PADE+U Dielectric"
@@ -431,6 +464,7 @@ class CalcType(ValueEnum):
     RevPBE_PADE_U_Static = "RevPBE+PADE+U Static"
     RevPBE_PADE_U_Structure_Optimization = "RevPBE+PADE+U Structure Optimization"
     RevPBE_PADE_U_Deformation = "RevPBE+PADE+U Deformation"
+    RevPBE_PADE_U_Unrecognized = "RevPBE+PADE+U Unrecognized"
     optB86b_U_NSCF_Line = "optB86b+U NSCF Line"
     optB86b_U_NSCF_Uniform = "optB86b+U NSCF Uniform"
     optB86b_U_Dielectric = "optB86b+U Dielectric"
@@ -441,6 +475,7 @@ class CalcType(ValueEnum):
     optB86b_U_Static = "optB86b+U Static"
     optB86b_U_Structure_Optimization = "optB86b+U Structure Optimization"
     optB86b_U_Deformation = "optB86b+U Deformation"
+    optB86b_U_Unrecognized = "optB86b+U Unrecognized"
     optB88_U_NSCF_Line = "optB88+U NSCF Line"
     optB88_U_NSCF_Uniform = "optB88+U NSCF Uniform"
     optB88_U_Dielectric = "optB88+U Dielectric"
@@ -451,6 +486,7 @@ class CalcType(ValueEnum):
     optB88_U_Static = "optB88+U Static"
     optB88_U_Structure_Optimization = "optB88+U Structure Optimization"
     optB88_U_Deformation = "optB88+U Deformation"
+    optB88_U_Unrecognized = "optB88+U Unrecognized"
     optPBE_U_NSCF_Line = "optPBE+U NSCF Line"
     optPBE_U_NSCF_Uniform = "optPBE+U NSCF Uniform"
     optPBE_U_Dielectric = "optPBE+U Dielectric"
@@ -461,6 +497,7 @@ class CalcType(ValueEnum):
     optPBE_U_Static = "optPBE+U Static"
     optPBE_U_Structure_Optimization = "optPBE+U Structure Optimization"
     optPBE_U_Deformation = "optPBE+U Deformation"
+    optPBE_U_Unrecognized = "optPBE+U Unrecognized"
     revPBE_U_NSCF_Line = "revPBE+U NSCF Line"
     revPBE_U_NSCF_Uniform = "revPBE+U NSCF Uniform"
     revPBE_U_Dielectric = "revPBE+U Dielectric"
@@ -471,6 +508,7 @@ class CalcType(ValueEnum):
     revPBE_U_Static = "revPBE+U Static"
     revPBE_U_Structure_Optimization = "revPBE+U Structure Optimization"
     revPBE_U_Deformation = "revPBE+U Deformation"
+    revPBE_U_Unrecognized = "revPBE+U Unrecognized"
     B3LYP_U_NSCF_Line = "B3LYP+U NSCF Line"
     B3LYP_U_NSCF_Uniform = "B3LYP+U NSCF Uniform"
     B3LYP_U_Dielectric = "B3LYP+U Dielectric"
@@ -481,6 +519,7 @@ class CalcType(ValueEnum):
     B3LYP_U_Static = "B3LYP+U Static"
     B3LYP_U_Structure_Optimization = "B3LYP+U Structure Optimization"
     B3LYP_U_Deformation = "B3LYP+U Deformation"
+    B3LYP_U_Unrecognized = "B3LYP+U Unrecognized"
     HF_U_NSCF_Line = "HF+U NSCF Line"
     HF_U_NSCF_Uniform = "HF+U NSCF Uniform"
     HF_U_Dielectric = "HF+U Dielectric"
@@ -491,6 +530,7 @@ class CalcType(ValueEnum):
     HF_U_Static = "HF+U Static"
     HF_U_Structure_Optimization = "HF+U Structure Optimization"
     HF_U_Deformation = "HF+U Deformation"
+    HF_U_Unrecognized = "HF+U Unrecognized"
     HSE03_U_NSCF_Line = "HSE03+U NSCF Line"
     HSE03_U_NSCF_Uniform = "HSE03+U NSCF Uniform"
     HSE03_U_Dielectric = "HSE03+U Dielectric"
@@ -501,6 +541,7 @@ class CalcType(ValueEnum):
     HSE03_U_Static = "HSE03+U Static"
     HSE03_U_Structure_Optimization = "HSE03+U Structure Optimization"
     HSE03_U_Deformation = "HSE03+U Deformation"
+    HSE03_U_Unrecognized = "HSE03+U Unrecognized"
     HSE06_U_NSCF_Line = "HSE06+U NSCF Line"
     HSE06_U_NSCF_Uniform = "HSE06+U NSCF Uniform"
     HSE06_U_Dielectric = "HSE06+U Dielectric"
@@ -511,6 +552,7 @@ class CalcType(ValueEnum):
     HSE06_U_Static = "HSE06+U Static"
     HSE06_U_Structure_Optimization = "HSE06+U Structure Optimization"
     HSE06_U_Deformation = "HSE06+U Deformation"
+    HSE06_U_Unrecognized = "HSE06+U Unrecognized"
     PB0_U_NSCF_Line = "PB0+U NSCF Line"
     PB0_U_NSCF_Uniform = "PB0+U NSCF Uniform"
     PB0_U_Dielectric = "PB0+U Dielectric"
@@ -521,6 +563,7 @@ class CalcType(ValueEnum):
     PB0_U_Static = "PB0+U Static"
     PB0_U_Structure_Optimization = "PB0+U Structure Optimization"
     PB0_U_Deformation = "PB0+U Deformation"
+    PB0_U_Unrecognized = "PB0+U Unrecognized"
     M06L_U_NSCF_Line = "M06L+U NSCF Line"
     M06L_U_NSCF_Uniform = "M06L+U NSCF Uniform"
     M06L_U_Dielectric = "M06L+U Dielectric"
@@ -531,6 +574,7 @@ class CalcType(ValueEnum):
     M06L_U_Static = "M06L+U Static"
     M06L_U_Structure_Optimization = "M06L+U Structure Optimization"
     M06L_U_Deformation = "M06L+U Deformation"
+    M06L_U_Unrecognized = "M06L+U Unrecognized"
     MBJL_U_NSCF_Line = "MBJL+U NSCF Line"
     MBJL_U_NSCF_Uniform = "MBJL+U NSCF Uniform"
     MBJL_U_Dielectric = "MBJL+U Dielectric"
@@ -541,6 +585,7 @@ class CalcType(ValueEnum):
     MBJL_U_Static = "MBJL+U Static"
     MBJL_U_Structure_Optimization = "MBJL+U Structure Optimization"
     MBJL_U_Deformation = "MBJL+U Deformation"
+    MBJL_U_Unrecognized = "MBJL+U Unrecognized"
     MS0_U_NSCF_Line = "MS0+U NSCF Line"
     MS0_U_NSCF_Uniform = "MS0+U NSCF Uniform"
     MS0_U_Dielectric = "MS0+U Dielectric"
@@ -551,6 +596,7 @@ class CalcType(ValueEnum):
     MS0_U_Static = "MS0+U Static"
     MS0_U_Structure_Optimization = "MS0+U Structure Optimization"
     MS0_U_Deformation = "MS0+U Deformation"
+    MS0_U_Unrecognized = "MS0+U Unrecognized"
     MS1_U_NSCF_Line = "MS1+U NSCF Line"
     MS1_U_NSCF_Uniform = "MS1+U NSCF Uniform"
     MS1_U_Dielectric = "MS1+U Dielectric"
@@ -561,6 +607,7 @@ class CalcType(ValueEnum):
     MS1_U_Static = "MS1+U Static"
     MS1_U_Structure_Optimization = "MS1+U Structure Optimization"
     MS1_U_Deformation = "MS1+U Deformation"
+    MS1_U_Unrecognized = "MS1+U Unrecognized"
     MS2_U_NSCF_Line = "MS2+U NSCF Line"
     MS2_U_NSCF_Uniform = "MS2+U NSCF Uniform"
     MS2_U_Dielectric = "MS2+U Dielectric"
@@ -571,6 +618,7 @@ class CalcType(ValueEnum):
     MS2_U_Static = "MS2+U Static"
     MS2_U_Structure_Optimization = "MS2+U Structure Optimization"
     MS2_U_Deformation = "MS2+U Deformation"
+    MS2_U_Unrecognized = "MS2+U Unrecognized"
     RTPSS_U_NSCF_Line = "RTPSS+U NSCF Line"
     RTPSS_U_NSCF_Uniform = "RTPSS+U NSCF Uniform"
     RTPSS_U_Dielectric = "RTPSS+U Dielectric"
@@ -581,6 +629,7 @@ class CalcType(ValueEnum):
     RTPSS_U_Static = "RTPSS+U Static"
     RTPSS_U_Structure_Optimization = "RTPSS+U Structure Optimization"
     RTPSS_U_Deformation = "RTPSS+U Deformation"
+    RTPSS_U_Unrecognized = "RTPSS+U Unrecognized"
     SCAN_U_NSCF_Line = "SCAN+U NSCF Line"
     SCAN_U_NSCF_Uniform = "SCAN+U NSCF Uniform"
     SCAN_U_Dielectric = "SCAN+U Dielectric"
@@ -591,6 +640,7 @@ class CalcType(ValueEnum):
     SCAN_U_Static = "SCAN+U Static"
     SCAN_U_Structure_Optimization = "SCAN+U Structure Optimization"
     SCAN_U_Deformation = "SCAN+U Deformation"
+    SCAN_U_Unrecognized = "SCAN+U Unrecognized"
     TPSS_U_NSCF_Line = "TPSS+U NSCF Line"
     TPSS_U_NSCF_Uniform = "TPSS+U NSCF Uniform"
     TPSS_U_Dielectric = "TPSS+U Dielectric"
@@ -601,6 +651,7 @@ class CalcType(ValueEnum):
     TPSS_U_Static = "TPSS+U Static"
     TPSS_U_Structure_Optimization = "TPSS+U Structure Optimization"
     TPSS_U_Deformation = "TPSS+U Deformation"
+    TPSS_U_Unrecognized = "TPSS+U Unrecognized"
     SCAN_rVV10_U_NSCF_Line = "SCAN-rVV10+U NSCF Line"
     SCAN_rVV10_U_NSCF_Uniform = "SCAN-rVV10+U NSCF Uniform"
     SCAN_rVV10_U_Dielectric = "SCAN-rVV10+U Dielectric"
@@ -613,6 +664,7 @@ class CalcType(ValueEnum):
     SCAN_rVV10_U_Static = "SCAN-rVV10+U Static"
     SCAN_rVV10_U_Structure_Optimization = "SCAN-rVV10+U Structure Optimization"
     SCAN_rVV10_U_Deformation = "SCAN-rVV10+U Deformation"
+    SCAN_rVV10_U_Unrecognized = "SCAN-rVV10+U Unrecognized"
     optB86b_vdW_U_NSCF_Line = "optB86b-vdW+U NSCF Line"
     optB86b_vdW_U_NSCF_Uniform = "optB86b-vdW+U NSCF Uniform"
     optB86b_vdW_U_Dielectric = "optB86b-vdW+U Dielectric"
@@ -625,6 +677,7 @@ class CalcType(ValueEnum):
     optB86b_vdW_U_Static = "optB86b-vdW+U Static"
     optB86b_vdW_U_Structure_Optimization = "optB86b-vdW+U Structure Optimization"
     optB86b_vdW_U_Deformation = "optB86b-vdW+U Deformation"
+    optB86b_vdW_U_Unrecognized = "optB86b-vdW+U Unrecognized"
     optB88_vdW_U_NSCF_Line = "optB88-vdW+U NSCF Line"
     optB88_vdW_U_NSCF_Uniform = "optB88-vdW+U NSCF Uniform"
     optB88_vdW_U_Dielectric = "optB88-vdW+U Dielectric"
@@ -637,6 +690,7 @@ class CalcType(ValueEnum):
     optB88_vdW_U_Static = "optB88-vdW+U Static"
     optB88_vdW_U_Structure_Optimization = "optB88-vdW+U Structure Optimization"
     optB88_vdW_U_Deformation = "optB88-vdW+U Deformation"
+    optB88_vdW_U_Unrecognized = "optB88-vdW+U Unrecognized"
     optPBE_vdW_U_NSCF_Line = "optPBE-vdW+U NSCF Line"
     optPBE_vdW_U_NSCF_Uniform = "optPBE-vdW+U NSCF Uniform"
     optPBE_vdW_U_Dielectric = "optPBE-vdW+U Dielectric"
@@ -649,6 +703,7 @@ class CalcType(ValueEnum):
     optPBE_vdW_U_Static = "optPBE-vdW+U Static"
     optPBE_vdW_U_Structure_Optimization = "optPBE-vdW+U Structure Optimization"
     optPBE_vdW_U_Deformation = "optPBE-vdW+U Deformation"
+    optPBE_vdW_U_Unrecognized = "optPBE-vdW+U Unrecognized"
     rev_vdW_DF2_U_NSCF_Line = "rev-vdW-DF2+U NSCF Line"
     rev_vdW_DF2_U_NSCF_Uniform = "rev-vdW-DF2+U NSCF Uniform"
     rev_vdW_DF2_U_Dielectric = "rev-vdW-DF2+U Dielectric"
@@ -661,6 +716,7 @@ class CalcType(ValueEnum):
     rev_vdW_DF2_U_Static = "rev-vdW-DF2+U Static"
     rev_vdW_DF2_U_Structure_Optimization = "rev-vdW-DF2+U Structure Optimization"
     rev_vdW_DF2_U_Deformation = "rev-vdW-DF2+U Deformation"
+    rev_vdW_DF2_U_Unrecognized = "rev-vdW-DF2+U Unrecognized"
     revPBE_vdW_U_NSCF_Line = "revPBE-vdW+U NSCF Line"
     revPBE_vdW_U_NSCF_Uniform = "revPBE-vdW+U NSCF Uniform"
     revPBE_vdW_U_Dielectric = "revPBE-vdW+U Dielectric"
@@ -673,6 +729,7 @@ class CalcType(ValueEnum):
     revPBE_vdW_U_Static = "revPBE-vdW+U Static"
     revPBE_vdW_U_Structure_Optimization = "revPBE-vdW+U Structure Optimization"
     revPBE_vdW_U_Deformation = "revPBE-vdW+U Deformation"
+    revPBE_vdW_U_Unrecognized = "revPBE-vdW+U Unrecognized"
     vdW_DF2_U_NSCF_Line = "vdW-DF2+U NSCF Line"
     vdW_DF2_U_NSCF_Uniform = "vdW-DF2+U NSCF Uniform"
     vdW_DF2_U_Dielectric = "vdW-DF2+U Dielectric"
@@ -683,6 +740,7 @@ class CalcType(ValueEnum):
     vdW_DF2_U_Static = "vdW-DF2+U Static"
     vdW_DF2_U_Structure_Optimization = "vdW-DF2+U Structure Optimization"
     vdW_DF2_U_Deformation = "vdW-DF2+U Deformation"
+    vdW_DF2_U_Unrecognized = "vdW-DF2+U Unrecognized"
     LDA_NSCF_Line = "LDA NSCF Line"
     LDA_NSCF_Uniform = "LDA NSCF Uniform"
     LDA_Dielectric = "LDA Dielectric"
@@ -693,6 +751,7 @@ class CalcType(ValueEnum):
     LDA_Static = "LDA Static"
     LDA_Structure_Optimization = "LDA Structure Optimization"
     LDA_Deformation = "LDA Deformation"
+    LDA_Unrecognized = "LDA Unrecognized"
     LDA_U_NSCF_Line = "LDA+U NSCF Line"
     LDA_U_NSCF_Uniform = "LDA+U NSCF Uniform"
     LDA_U_Dielectric = "LDA+U Dielectric"
@@ -703,3 +762,4 @@ class CalcType(ValueEnum):
     LDA_U_Static = "LDA+U Static"
     LDA_U_Structure_Optimization = "LDA+U Structure Optimization"
     LDA_U_Deformation = "LDA+U Deformation"
+    LDA_U_Unrecognized = "LDA+U Unrecognized"
diff --git a/emmet-core/emmet/core/vasp/calc_types/generate.py b/emmet-core/emmet/core/vasp/calc_types/generate.py
index 825df77d62..16d0f1f3bd 100644
--- a/emmet-core/emmet/core/vasp/calc_types/generate.py
+++ b/emmet-core/emmet/core/vasp/calc_types/generate.py
@@ -1,20 +1,8 @@
 """ Module to define various calculation types as Enums for VASP """
-import datetime
-from itertools import groupby, product
+from itertools import product
 from pathlib import Path
-from typing import Dict, Iterator, List
 
-import bson
-import numpy as np
-from monty.json import MSONable
 from monty.serialization import loadfn
-from pydantic import BaseModel
-from pymatgen.analysis.structure_matcher import ElementComparator, StructureMatcher
-from pymatgen.core.structure import Structure
-from typing_extensions import Literal
-
-from emmet.core import SETTINGS
-from emmet.core.utils import ValueEnum
 
 _RUN_TYPE_DATA = loadfn(str(Path(__file__).parent.joinpath("run_types.yaml").resolve()))
 _TASK_TYPES = [
@@ -28,6 +16,7 @@
     "Static",
     "Structure Optimization",
     "Deformation",
+    "Unrecognized",
 ]
 
 _RUN_TYPES = (

From 03d3512cc6b2f512faa25a1a1a245ef87eee7bf9 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Fri, 28 May 2021 14:01:43 -0700
Subject: [PATCH 23/26] fix minor conversion bugs

---
 emmet-builders/emmet/builders/vasp/thermo.py | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/emmet-builders/emmet/builders/vasp/thermo.py b/emmet-builders/emmet/builders/vasp/thermo.py
index c237ff2b2f..a64616a934 100644
--- a/emmet-builders/emmet/builders/vasp/thermo.py
+++ b/emmet-builders/emmet/builders/vasp/thermo.py
@@ -224,9 +224,9 @@ def get_entries(self, chemsys: str) -> List[Dict]:
         if self.oxidation_states:
             material_ids = [t["material_id"] for t in materials_docs]
             oxi_states_data = {
-                d["material_id"]: d["bond_valence"]["average_oxidation_states"]
+                d["material_id"]: d.get("average_oxidation_states", {})
                 for d in self.oxidation_states.query(
-                    properties=["material_id", "bond_valence.average_oxidation_states"],
+                    properties=["material_id", "average_oxidation_states"],
                     criteria={"material_id": {"$in": material_ids}, "successful": True},
                 )
             }
@@ -258,7 +258,7 @@ def get_updated_chemsys(
         updated_mats = self.thermo.newer_in(self.materials, criteria=self.query)
         updated_chemsys = set(
             self.materials.distinct(
-                "chemsys", {self.materials.key: {"$in": list(updated_mats)}}
+                "chemsys", {"material_id": {"$in": list(updated_mats)}}
             )
         )
         self.logger.debug(f"Found {len(updated_chemsys)} updated chemical systems")
@@ -269,8 +269,8 @@ def get_new_chemsys(self) -> Set:
         """Gets newer chemical system as defined by introduction of a new material"""
 
         # All materials that are not present in the thermo collection
-        thermo_mat_ids = self.thermo.distinct(self.thermo.key)
-        mat_ids = self.materials.distinct(self.materials.key, self.query)
+        thermo_mat_ids = self.thermo.distinct("material_id")
+        mat_ids = self.materials.distinct("material_id", self.query)
         dif_task_ids = list(set(mat_ids) - set(thermo_mat_ids))
         q = {"material_id": {"$in": dif_task_ids}}
         new_mat_chemsys = set(self.materials.distinct("chemsys", q))

From 2994cb4cb74ec7b381bae21d415dc3808c63e0eb Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Fri, 28 May 2021 14:14:25 -0700
Subject: [PATCH 24/26] fix counting of chemsys to process

---
 emmet-builders/emmet/builders/vasp/thermo.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/emmet-builders/emmet/builders/vasp/thermo.py b/emmet-builders/emmet/builders/vasp/thermo.py
index a64616a934..1b971e64c2 100644
--- a/emmet-builders/emmet/builders/vasp/thermo.py
+++ b/emmet-builders/emmet/builders/vasp/thermo.py
@@ -62,6 +62,7 @@ def ensure_indexes(self):
 
         # Search index for materials
         self.materials.ensure_index("material_id")
+        self.materials.ensure_index("chemsys")
         self.materials.ensure_index("last_updated")
 
         # Search index for thermo
@@ -99,10 +100,12 @@ def get_items(self) -> Iterator[List[Dict]]:
         affected_chemsys = self.get_affected_chemsys(updated_chemsys | new_chemsys)
 
         # Remove overlapping chemical systems
-        to_process_chemsys = set()
+        processed = set()
+        to_process_chemsys = []
         for chemsys in updated_chemsys | new_chemsys | affected_chemsys:
-            if chemsys not in to_process_chemsys:
-                to_process_chemsys |= chemsys_permutations(chemsys)
+            if chemsys not in processed:
+                processed |= chemsys_permutations(chemsys)
+                to_process_chemsys.append(chemsys)
 
         self.logger.info(
             f"Found {len(to_process_chemsys)} chemical systems with new/updated materials to process"

From fabbe34dc3796bc3a7c257fd2d449d25bcec4162 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Fri, 28 May 2021 14:16:51 -0700
Subject: [PATCH 25/26] catch warnings from compatability

---
 emmet-builders/emmet/builders/vasp/thermo.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/emmet-builders/emmet/builders/vasp/thermo.py b/emmet-builders/emmet/builders/vasp/thermo.py
index 1b971e64c2..591f8aaac6 100644
--- a/emmet-builders/emmet/builders/vasp/thermo.py
+++ b/emmet-builders/emmet/builders/vasp/thermo.py
@@ -1,3 +1,4 @@
+import warnings
 from collections import defaultdict
 from itertools import chain
 from typing import Dict, Iterable, Iterator, List, Optional, Set, Tuple
@@ -137,7 +138,11 @@ def process_item(self, item: List[Dict]):
         for entry in entries:
             material_entries[entry.entry_id][entry.data["run_type"]] = entry
 
-        pd_entries = self.compatibility.process_entries(entries)
+        with warnings.catch_warnings():
+            warnings.filterwarnings(
+                "ignore", message="Failed to guess oxidation states.*"
+            )
+            pd_entries = self.compatibility.process_entries(entries)
         self.logger.debug(f"{len(pd_entries)} remain in {chemsys} after filtering")
 
         try:

From cb293cd108201555c0595638a94b336d7c9925e9 Mon Sep 17 00:00:00 2001
From: Shyam D <shyamd@lbl.gov>
Date: Fri, 28 May 2021 15:43:30 -0700
Subject: [PATCH 26/26] fix bad structures into oxidation doc test

---
 tests/emmet-core/test_oxidation_states.py | 23 +++--------------------
 1 file changed, 3 insertions(+), 20 deletions(-)

diff --git a/tests/emmet-core/test_oxidation_states.py b/tests/emmet-core/test_oxidation_states.py
index df19ae60bf..7c73fbac71 100644
--- a/tests/emmet-core/test_oxidation_states.py
+++ b/tests/emmet-core/test_oxidation_states.py
@@ -1,3 +1,4 @@
+import pydantic
 import pytest
 from pymatgen.core import Structure
 from pymatgen.util.testing import PymatgenTest
@@ -5,7 +6,7 @@
 from emmet.core.oxidation_states import OxidationStateDoc
 
 test_structures = {
-    name: struc
+    name: struc.get_reduced_structure()
     for name, struc in PymatgenTest.TEST_STRUCTURES.items()
     if name
     in [
@@ -20,14 +21,6 @@
         "NaFePO4",
         "Pb2TiZrO6",
         "SrTiO3",
-    ]
-}
-
-fail_structures = {
-    name: struc
-    for name, struc in PymatgenTest.TEST_STRUCTURES.items()
-    if name
-    in [
         "TiO2",
         "BaNiO3",
         "VO2",
@@ -38,16 +31,6 @@
 @pytest.mark.parametrize("structure", test_structures.values())
 def test_oxidation_state(structure: Structure):
     """Very simple test to make sure this actually works"""
-
+    print(f"Should work : {structure.composition}")
     doc = OxidationStateDoc.from_structure(structure)
-    print(structure.composition)
     assert doc is not None
-
-
-@pytest.mark.parametrize("structure", fail_structures.values())
-def test_oxidation_state_failures(structure: Structure):
-    """Very simple test to make sure this actually fails"""
-
-    doc = OxidationStateDoc.from_structure(structure)
-    print(structure.composition)
-    assert doc is None